def evaluate(individual):
keys = list(parameters.keys())
dim = len(keys)
# print(type(individual))
# print(individual)
assert len(individual) == dim
for i in range(dim):
(lo, up) = para_bounds[i]
if individual[i] < lo:
sentence = 'increase ' + str(keys[i]) + ' : ' + str(
individual[i]) + ' to lower bound !'
print(sentence)
individual[i] = lo
elif individual[i] > up:
sentence = 'decrease ' + str(keys[i]) + ' : ' + str(
individual[i]) + ' to upper bound !'
print(sentence)
individual[i] = up
BayesOp_Parameters = dict(zip(keys, individual))
BayesOp_Parameters['silent'] = True
BayesOp_Parameters['nthread'] = -1
BayesOp_Parameters['seed'] = 0
# BayesOp_Parameters['objective'] = "multi:softprob"
BayesOp_Parameters['max_depth'] = int(BayesOp_Parameters['max_depth'])
BayesOp_Parameters['n_estimators'] = int(
BayesOp_Parameters['n_estimators'])
'''
r = np.load(name)
Feature_train_list = r['F_tr_l']
Label_train_list = r['L_tr_l']
Feature_valid_list = r['F_va_l']
Label_valid_list = r['L_va_l']
'''
Feature_train_list = data_dic['F_tr_l']
Label_train_list = data_dic['L_tr_l']
Feature_valid_list = data_dic['F_va_l']
Label_valid_list = data_dic['L_va_l']
Num_list = len(Feature_train_list)
Num_Cross_Folders = Num_list
ml_record = MetricList(Num_Cross_Folders)
i = 0
for j in range(Num_list):
Feature_train = Feature_train_list[j]
Label_train = Label_train_list[j]
Feature_valid = Feature_valid_list[j]
Label_valid = Label_valid_list[j]
Label_valid.ravel()
xgb = xgboost.XGBClassifier(**BayesOp_Parameters)
xgb.fit(Feature_train, Label_train.ravel())
Label_predict = xgb.predict(Feature_valid)
ml_record.measure(i, Label_valid, Label_predict, 'weighted')
i += 1
return ml_record.mean_G(),
def xgboostcv(
max_depth,
learning_rate,
n_estimators,
gamma,
min_child_weight,
max_delta_step,
subsample,
colsample_bytree,
silent=True,
nthread=-1,
seed=0,
):
r = np.load(name)
Feature_train_list = r['F_tr_l']
Label_train_list = r['L_tr_l']
Feature_valid_list = r['F_va_l']
Label_valid_list = r['L_va_l']
Num_list = len(Feature_train_list)
Num_Cross_Folders = Num_list
ml_record = MetricList(Num_Cross_Folders)
i = 0
for j in range(Num_list):
Feature_train = Feature_train_list[j]
Label_train = Label_train_list[j]
# Num_train = Feature_train_list[j].shape[0]
Feature_valid = Feature_valid_list[j]
Label_valid = Label_valid_list[j]
Label_valid.ravel()
# Num_valid = Feature_valid_list[j].shape[0]
xgb = xgboost.XGBClassifier(max_depth=int(max_depth),
learning_rate=learning_rate,
n_estimators=int(n_estimators),
silent=silent,
nthread=nthread,
gamma=gamma,
min_child_weight=min_child_weight,
max_delta_step=max_delta_step,
subsample=subsample,
colsample_bytree=colsample_bytree,
seed=seed,
objective="multi:softprob")
xgb.fit(Feature_train, Label_train.ravel())
Label_predict = xgb.predict(Feature_valid)
ml_record.measure(i, Label_valid, Label_predict, 'weighted')
i += 1
return ml_record.mean_G()
def evaluate(individual):
keys = list(parameters.keys())
dim = len(keys)
# print(type(individual))
# print(individual)
assert len(individual) == dim
for i in range(dim):
(lo, up) = para_bounds[i]
if individual[i] < lo:
sentence = 'increase ' + str(keys[i]) + ' : ' + str(
individual[i]) + ' to lower bound !'
print(sentence)
individual[i] = lo
elif individual[i] > up:
sentence = 'decrease ' + str(keys[i]) + ' : ' + str(
individual[i]) + ' to upper bound !'
print(sentence)
individual[i] = up
BayesOp_Parameters = dict(zip(keys, individual))
BayesOp_Parameters['silent'] = True
BayesOp_Parameters['nthread'] = -1
BayesOp_Parameters['seed'] = 1234
# BayesOp_Parameters['objective'] = "multi:softprob"
BayesOp_Parameters['max_depth'] = int(BayesOp_Parameters['max_depth'])
BayesOp_Parameters['n_estimators'] = int(
BayesOp_Parameters['n_estimators'])
Num_Cross_Folders = 5
ml_record = MetricList(Num_Cross_Folders)
i = 0
for file in files:
name = dir_path + '/' + file
r = np.load(name)
Feature_train = r['F_tr']
Label_train = r['L_tr']
Num_train = Feature_train.shape[0]
# print(Num_train)
Feature_test = r['F_te']
Label_test = r['L_te']
Label_test.ravel()
Num_test = Feature_test.shape[0]
# print(Num_test)
xgb = xgboost.XGBClassifier(**BayesOp_Parameters)
xgb.fit(Feature_train, Label_train.ravel())
Label_predict = xgb.predict(Feature_test)
ml_record.measure(i, Label_test, Label_predict, 'weighted')
i += 1
return ml_record.mean_G(),
def xgboostcv(max_depth,
learning_rate,
n_estimators,
gamma,
min_child_weight,
max_delta_step,
subsample,
colsample_bytree,
silent =True,
nthread = -1,
seed = 1234,
):
Num_Cross_Folders = 5
ml_record = MetricList(Num_Cross_Folders)
i = 0
for file in files:
name = dir_path + '/' + file
r = np.load(name)
Feature_train = r['F_tr']
Label_train = r['L_tr']
Num_train = Feature_train.shape[0]
# print(Num_train)
Feature_test = r['F_te']
Label_test = r['L_te']
Label_test.ravel()
Num_test = Feature_test.shape[0]
# print(Num_test)
xgb = xgboost.XGBClassifier(max_depth = int(max_depth),
learning_rate = learning_rate,
n_estimators = int(n_estimators),
silent = silent,
nthread = nthread,
gamma = gamma,
min_child_weight = min_child_weight,
max_delta_step = max_delta_step,
subsample = subsample,
colsample_bytree = colsample_bytree,
seed = seed,
objective = "multi:softprob")
xgb.fit(Feature_train, Label_train.ravel())
Label_predict = xgb.predict(Feature_test)
ml_record.measure(i, Label_test, Label_predict, 'weighted')
i += 1
return ml_record.mean_G()
def evaluate(para_value):
keys = para_keys
dim = len(keys)
BayesOp_Parameters = dict(zip(keys, para_value))
BayesOp_Parameters['silent'] = True
BayesOp_Parameters['nthread'] = -1
BayesOp_Parameters['seed'] = 0
# BayesOp_Parameters['objective'] = "multi:softprob"
BayesOp_Parameters['max_depth'] = int(BayesOp_Parameters['max_depth'])
BayesOp_Parameters['n_estimators'] = int(BayesOp_Parameters['n_estimators'])
Num_Cross_Folders = 5
ml_record = MetricList(Num_Cross_Folders)
i = 0
for file in files:
name = dir_path + '/' + file
r = np.load(name)
Feature_train = r['F_tr']
Label_train = r['L_tr']
Num_train = Feature_train.shape[0]
# print(Num_train)
Feature_test = r['F_te']
Label_test = r['L_te']
Label_test.ravel()
Num_test = Feature_test.shape[0]
# print(Num_test)
xgb = xgboost.XGBClassifier(**BayesOp_Parameters)
xgb.fit(Feature_train, Label_train.ravel())
Label_predict = xgb.predict(Feature_test)
ml_record.measure(i, Label_test, Label_predict, 'weighted')
i += 1
return 1-ml_record.mean_G()
from imblearn.over_sampling import SMOTE
from metrics_list import MetricList
path = "KEEL_Cross_Folder_npz"
dirs = os.listdir(path) #Get files in the folder
for Dir in dirs:
print("Data Set Name: ", Dir)
dir_path = path + "/" + Dir
files = os.listdir(dir_path) # Get files in the folder
methods = ["XGBoost", "SMOTE-XGBoost", "AdaBoost-DT", "SMOTE-AdaBoost-DT", "META-DES", "MCB", "DES-MI", "One_vs_Rest-SMOTE-XGBoost", "One_vs_Rest-XGBoost"]
for m in methods:
print('Method: ', m)
Num_Cross_Folders = 5
ml_record = MetricList(Num_Cross_Folders)
i = 0
for file in files:
name = dir_path + '/' + file
r = np.load(name)
Feature_train = r['F_tr']
Label_train = r['L_tr']
Num_train = Feature_train.shape[0]
#print(Num_train)
Feature_test = r['F_te']
Label_test = r['L_te']
Label_test.ravel()
Num_test = Feature_test.shape[0]
#print(Num_test)
print(i, " folder; ", "Number of train: ", Num_train, "Number of test: ", Num_test)
